blob: cc0e8f3f20b30642547d3fbeb1f0f426c5484c20 [file] [log] [blame]
/*
* Copyright (c) 2010-2014, 2016-2018 ARM Limited
* Copyright (c) 2013 Advanced Micro Devices, Inc.
* All rights reserved
*
* The license below extends only to copyright in the software and shall
* not be construed as granting a license to any other intellectual
* property including but not limited to intellectual property relating
* to a hardware implementation of the functionality of the software
* licensed hereunder. You may use the software subject to the license
* terms below provided that you ensure that this notice is replicated
* unmodified and in its entirety in all distributions of the software,
* modified or unmodified, in source code or in binary form.
*
* Copyright (c) 2007-2008 The Florida State University
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met: redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer;
* redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution;
* neither the name of the copyright holders nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* Authors: Stephen Hines
*/
#include "arch/arm/insts/static_inst.hh"
#include "arch/arm/faults.hh"
#include "arch/arm/isa.hh"
#include "base/condcodes.hh"
#include "base/cprintf.hh"
#include "base/loader/symtab.hh"
#include "cpu/reg_class.hh"
namespace ArmISA
{
// Shift Rm by an immediate value
int32_t
ArmStaticInst::shift_rm_imm(uint32_t base, uint32_t shamt,
uint32_t type, uint32_t cfval) const
{
assert(shamt < 32);
ArmShiftType shiftType;
shiftType = (ArmShiftType)type;
switch (shiftType)
{
case LSL:
return base << shamt;
case LSR:
if (shamt == 0)
return 0;
else
return base >> shamt;
case ASR:
if (shamt == 0)
return (base >> 31) | -((base & (1 << 31)) >> 31);
else
return (base >> shamt) | -((base & (1 << 31)) >> shamt);
case ROR:
if (shamt == 0)
return (cfval << 31) | (base >> 1); // RRX
else
return (base << (32 - shamt)) | (base >> shamt);
default:
ccprintf(std::cerr, "Unhandled shift type\n");
exit(1);
break;
}
return 0;
}
int64_t
ArmStaticInst::shiftReg64(uint64_t base, uint64_t shiftAmt,
ArmShiftType type, uint8_t width) const
{
shiftAmt = shiftAmt % width;
ArmShiftType shiftType;
shiftType = (ArmShiftType)type;
switch (shiftType)
{
case LSL:
return base << shiftAmt;
case LSR:
if (shiftAmt == 0)
return base;
else
return (base & mask(width)) >> shiftAmt;
case ASR:
if (shiftAmt == 0) {
return base;
} else {
int sign_bit = bits(base, intWidth - 1);
base >>= shiftAmt;
base = sign_bit ? (base | ~mask(intWidth - shiftAmt)) : base;
return base & mask(intWidth);
}
case ROR:
if (shiftAmt == 0)
return base;
else
return (base << (width - shiftAmt)) | (base >> shiftAmt);
default:
ccprintf(std::cerr, "Unhandled shift type\n");
exit(1);
break;
}
return 0;
}
int64_t
ArmStaticInst::extendReg64(uint64_t base, ArmExtendType type,
uint64_t shiftAmt, uint8_t width) const
{
bool sign_extend = false;
int len = 0;
switch (type) {
case UXTB:
len = 8;
break;
case UXTH:
len = 16;
break;
case UXTW:
len = 32;
break;
case UXTX:
len = 64;
break;
case SXTB:
len = 8;
sign_extend = true;
break;
case SXTH:
len = 16;
sign_extend = true;
break;
case SXTW:
len = 32;
sign_extend = true;
break;
case SXTX:
len = 64;
sign_extend = true;
break;
}
len = len <= width - shiftAmt ? len : width - shiftAmt;
uint64_t tmp = (uint64_t) bits(base, len - 1, 0) << shiftAmt;
if (sign_extend) {
int sign_bit = bits(tmp, len + shiftAmt - 1);
tmp = sign_bit ? (tmp | ~mask(len + shiftAmt)) : tmp;
}
return tmp & mask(width);
}
// Shift Rm by Rs
int32_t
ArmStaticInst::shift_rm_rs(uint32_t base, uint32_t shamt,
uint32_t type, uint32_t cfval) const
{
enum ArmShiftType shiftType;
shiftType = (enum ArmShiftType) type;
switch (shiftType)
{
case LSL:
if (shamt >= 32)
return 0;
else
return base << shamt;
case LSR:
if (shamt >= 32)
return 0;
else
return base >> shamt;
case ASR:
if (shamt >= 32)
return (base >> 31) | -((base & (1 << 31)) >> 31);
else
return (base >> shamt) | -((base & (1 << 31)) >> shamt);
case ROR:
shamt = shamt & 0x1f;
if (shamt == 0)
return base;
else
return (base << (32 - shamt)) | (base >> shamt);
default:
ccprintf(std::cerr, "Unhandled shift type\n");
exit(1);
break;
}
return 0;
}
// Generate C for a shift by immediate
bool
ArmStaticInst::shift_carry_imm(uint32_t base, uint32_t shamt,
uint32_t type, uint32_t cfval) const
{
enum ArmShiftType shiftType;
shiftType = (enum ArmShiftType) type;
switch (shiftType)
{
case LSL:
if (shamt == 0)
return cfval;
else
return (base >> (32 - shamt)) & 1;
case LSR:
if (shamt == 0)
return (base >> 31);
else
return (base >> (shamt - 1)) & 1;
case ASR:
if (shamt == 0)
return (base >> 31);
else
return (base >> (shamt - 1)) & 1;
case ROR:
shamt = shamt & 0x1f;
if (shamt == 0)
return (base & 1); // RRX
else
return (base >> (shamt - 1)) & 1;
default:
ccprintf(std::cerr, "Unhandled shift type\n");
exit(1);
break;
}
return 0;
}
// Generate C for a shift by Rs
bool
ArmStaticInst::shift_carry_rs(uint32_t base, uint32_t shamt,
uint32_t type, uint32_t cfval) const
{
enum ArmShiftType shiftType;
shiftType = (enum ArmShiftType) type;
if (shamt == 0)
return cfval;
switch (shiftType)
{
case LSL:
if (shamt > 32)
return 0;
else
return (base >> (32 - shamt)) & 1;
case LSR:
if (shamt > 32)
return 0;
else
return (base >> (shamt - 1)) & 1;
case ASR:
if (shamt > 32)
shamt = 32;
return (base >> (shamt - 1)) & 1;
case ROR:
shamt = shamt & 0x1f;
if (shamt == 0)
shamt = 32;
return (base >> (shamt - 1)) & 1;
default:
ccprintf(std::cerr, "Unhandled shift type\n");
exit(1);
break;
}
return 0;
}
void
ArmStaticInst::printIntReg(std::ostream &os, RegIndex reg_idx,
uint8_t opWidth) const
{
if (opWidth == 0)
opWidth = intWidth;
if (aarch64) {
if (reg_idx == INTREG_UREG0)
ccprintf(os, "ureg0");
else if (reg_idx == INTREG_SPX)
ccprintf(os, "%s%s", (opWidth == 32) ? "w" : "", "sp");
else if (reg_idx == INTREG_X31)
ccprintf(os, "%szr", (opWidth == 32) ? "w" : "x");
else
ccprintf(os, "%s%d", (opWidth == 32) ? "w" : "x", reg_idx);
} else {
switch (reg_idx) {
case PCReg:
ccprintf(os, "pc");
break;
case StackPointerReg:
ccprintf(os, "sp");
break;
case FramePointerReg:
ccprintf(os, "fp");
break;
case ReturnAddressReg:
ccprintf(os, "lr");
break;
default:
ccprintf(os, "r%d", reg_idx);
break;
}
}
}
void ArmStaticInst::printPFflags(std::ostream &os, int flag) const
{
const char *flagtoprfop[]= { "PLD", "PLI", "PST", "Reserved"};
const char *flagtotarget[] = { "L1", "L2", "L3", "Reserved"};
const char *flagtopolicy[] = { "KEEP", "STRM"};
ccprintf(os, "%s%s%s", flagtoprfop[(flag>>3)&3],
flagtotarget[(flag>>1)&3], flagtopolicy[flag&1]);
}
void
ArmStaticInst::printFloatReg(std::ostream &os, RegIndex reg_idx) const
{
ccprintf(os, "f%d", reg_idx);
}
void
ArmStaticInst::printVecReg(std::ostream &os, RegIndex reg_idx,
bool isSveVecReg) const
{
ccprintf(os, "%s%d", isSveVecReg ? "z" : "v", reg_idx);
}
void
ArmStaticInst::printVecPredReg(std::ostream &os, RegIndex reg_idx) const
{
ccprintf(os, "p%d", reg_idx);
}
void
ArmStaticInst::printCCReg(std::ostream &os, RegIndex reg_idx) const
{
ccprintf(os, "cc_%s", ArmISA::ccRegName[reg_idx]);
}
void
ArmStaticInst::printMiscReg(std::ostream &os, RegIndex reg_idx) const
{
assert(reg_idx < NUM_MISCREGS);
ccprintf(os, "%s", ArmISA::miscRegName[reg_idx]);
}
void
ArmStaticInst::printMnemonic(std::ostream &os,
const std::string &suffix,
bool withPred,
bool withCond64,
ConditionCode cond64) const
{
os << " " << mnemonic;
if (withPred && !aarch64) {
printCondition(os, machInst.condCode);
os << suffix;
} else if (withCond64) {
os << ".";
printCondition(os, cond64);
os << suffix;
}
if (machInst.bigThumb)
os << ".w";
os << " ";
}
void
ArmStaticInst::printTarget(std::ostream &os, Addr target,
const SymbolTable *symtab) const
{
Addr symbolAddr;
std::string symbol;
if (symtab && symtab->findNearestSymbol(target, symbol, symbolAddr)) {
ccprintf(os, "<%s", symbol);
if (symbolAddr != target)
ccprintf(os, "+%d>", target - symbolAddr);
else
ccprintf(os, ">");
} else {
ccprintf(os, "%#x", target);
}
}
void
ArmStaticInst::printCondition(std::ostream &os,
unsigned code,
bool noImplicit) const
{
switch (code) {
case COND_EQ:
os << "eq";
break;
case COND_NE:
os << "ne";
break;
case COND_CS:
os << "cs";
break;
case COND_CC:
os << "cc";
break;
case COND_MI:
os << "mi";
break;
case COND_PL:
os << "pl";
break;
case COND_VS:
os << "vs";
break;
case COND_VC:
os << "vc";
break;
case COND_HI:
os << "hi";
break;
case COND_LS:
os << "ls";
break;
case COND_GE:
os << "ge";
break;
case COND_LT:
os << "lt";
break;
case COND_GT:
os << "gt";
break;
case COND_LE:
os << "le";
break;
case COND_AL:
// This one is implicit.
if (noImplicit)
os << "al";
break;
case COND_UC:
// Unconditional.
if (noImplicit)
os << "uc";
break;
default:
panic("Unrecognized condition code %d.\n", code);
}
}
void
ArmStaticInst::printMemSymbol(std::ostream &os,
const SymbolTable *symtab,
const std::string &prefix,
const Addr addr,
const std::string &suffix) const
{
Addr symbolAddr;
std::string symbol;
if (symtab && symtab->findNearestSymbol(addr, symbol, symbolAddr)) {
ccprintf(os, "%s%s", prefix, symbol);
if (symbolAddr != addr)
ccprintf(os, "+%d", addr - symbolAddr);
ccprintf(os, suffix);
}
}
void
ArmStaticInst::printShiftOperand(std::ostream &os,
IntRegIndex rm,
bool immShift,
uint32_t shiftAmt,
IntRegIndex rs,
ArmShiftType type) const
{
bool firstOp = false;
if (rm != INTREG_ZERO) {
printIntReg(os, rm);
}
bool done = false;
if ((type == LSR || type == ASR) && immShift && shiftAmt == 0)
shiftAmt = 32;
switch (type) {
case LSL:
if (immShift && shiftAmt == 0) {
done = true;
break;
}
if (!firstOp)
os << ", ";
os << "LSL";
break;
case LSR:
if (!firstOp)
os << ", ";
os << "LSR";
break;
case ASR:
if (!firstOp)
os << ", ";
os << "ASR";
break;
case ROR:
if (immShift && shiftAmt == 0) {
if (!firstOp)
os << ", ";
os << "RRX";
done = true;
break;
}
if (!firstOp)
os << ", ";
os << "ROR";
break;
default:
panic("Tried to disassemble unrecognized shift type.\n");
}
if (!done) {
if (!firstOp)
os << " ";
if (immShift)
os << "#" << shiftAmt;
else
printIntReg(os, rs);
}
}
void
ArmStaticInst::printExtendOperand(bool firstOperand, std::ostream &os,
IntRegIndex rm, ArmExtendType type,
int64_t shiftAmt) const
{
if (!firstOperand)
ccprintf(os, ", ");
printIntReg(os, rm);
if (type == UXTX && shiftAmt == 0)
return;
switch (type) {
case UXTB: ccprintf(os, ", UXTB");
break;
case UXTH: ccprintf(os, ", UXTH");
break;
case UXTW: ccprintf(os, ", UXTW");
break;
case UXTX: ccprintf(os, ", LSL");
break;
case SXTB: ccprintf(os, ", SXTB");
break;
case SXTH: ccprintf(os, ", SXTH");
break;
case SXTW: ccprintf(os, ", SXTW");
break;
case SXTX: ccprintf(os, ", SXTW");
break;
}
if (type == UXTX || shiftAmt)
ccprintf(os, " #%d", shiftAmt);
}
void
ArmStaticInst::printDataInst(std::ostream &os, bool withImm,
bool immShift, bool s, IntRegIndex rd, IntRegIndex rn,
IntRegIndex rm, IntRegIndex rs, uint32_t shiftAmt,
ArmShiftType type, uint64_t imm) const
{
printMnemonic(os, s ? "s" : "");
bool firstOp = true;
// Destination
if (rd != INTREG_ZERO) {
firstOp = false;
printIntReg(os, rd);
}
// Source 1.
if (rn != INTREG_ZERO) {
if (!firstOp)
os << ", ";
firstOp = false;
printIntReg(os, rn);
}
if (!firstOp)
os << ", ";
if (withImm) {
ccprintf(os, "#%ld", imm);
} else {
printShiftOperand(os, rm, immShift, shiftAmt, rs, type);
}
}
std::string
ArmStaticInst::generateDisassembly(Addr pc,
const SymbolTable *symtab) const
{
std::stringstream ss;
printMnemonic(ss);
return ss.str();
}
Fault
ArmStaticInst::softwareBreakpoint32(ExecContext *xc, uint16_t imm) const
{
const auto tc = xc->tcBase();
const HCR hcr = tc->readMiscReg(MISCREG_HCR_EL2);
const HDCR mdcr = tc->readMiscRegNoEffect(MISCREG_MDCR_EL2);
if ((ArmSystem::haveEL(tc, EL2) && !inSecureState(tc) &&
!ELIs32(tc, EL2) && (hcr.tge == 1 || mdcr.tde == 1)) ||
!ELIs32(tc, EL1)) {
// Route to AArch64 Software Breakpoint
return std::make_shared<SoftwareBreakpoint>(machInst, imm);
} else {
// Execute AArch32 Software Breakpoint
return std::make_shared<PrefetchAbort>(readPC(xc),
ArmFault::DebugEvent);
}
}
Fault
ArmStaticInst::advSIMDFPAccessTrap64(ExceptionLevel el) const
{
switch (el) {
case EL1:
return std::make_shared<SupervisorTrap>(machInst, 0x1E00000,
EC_TRAPPED_SIMD_FP);
case EL2:
return std::make_shared<HypervisorTrap>(machInst, 0x1E00000,
EC_TRAPPED_SIMD_FP);
case EL3:
return std::make_shared<SecureMonitorTrap>(machInst, 0x1E00000,
EC_TRAPPED_SIMD_FP);
default:
panic("Illegal EL in advSIMDFPAccessTrap64\n");
}
}
Fault
ArmStaticInst::checkFPAdvSIMDTrap64(ThreadContext *tc, CPSR cpsr) const
{
if (ArmSystem::haveVirtualization(tc) && !inSecureState(tc)) {
HCPTR cptrEnCheck = tc->readMiscReg(MISCREG_CPTR_EL2);
if (cptrEnCheck.tfp)
return advSIMDFPAccessTrap64(EL2);
}
if (ArmSystem::haveSecurity(tc)) {
HCPTR cptrEnCheck = tc->readMiscReg(MISCREG_CPTR_EL3);
if (cptrEnCheck.tfp)
return advSIMDFPAccessTrap64(EL3);
}
return NoFault;
}
Fault
ArmStaticInst::checkFPAdvSIMDEnabled64(ThreadContext *tc,
CPSR cpsr, CPACR cpacr) const
{
const ExceptionLevel el = (ExceptionLevel) (uint8_t)cpsr.el;
if ((el == EL0 && cpacr.fpen != 0x3) ||
(el == EL1 && !(cpacr.fpen & 0x1)))
return advSIMDFPAccessTrap64(EL1);
return checkFPAdvSIMDTrap64(tc, cpsr);
}
Fault
ArmStaticInst::checkAdvSIMDOrFPEnabled32(ThreadContext *tc,
CPSR cpsr, CPACR cpacr,
NSACR nsacr, FPEXC fpexc,
bool fpexc_check, bool advsimd) const
{
const bool have_virtualization = ArmSystem::haveVirtualization(tc);
const bool have_security = ArmSystem::haveSecurity(tc);
const bool is_secure = inSecureState(tc);
const ExceptionLevel cur_el = opModeToEL(currOpMode(tc));
if (cur_el == EL0 && ELIs64(tc, EL1))
return checkFPAdvSIMDEnabled64(tc, cpsr, cpacr);
uint8_t cpacr_cp10 = cpacr.cp10;
bool cpacr_asedis = cpacr.asedis;
if (have_security && !ELIs64(tc, EL3) && !is_secure) {
if (nsacr.nsasedis)
cpacr_asedis = true;
if (nsacr.cp10 == 0)
cpacr_cp10 = 0;
}
if (cur_el != EL2) {
if (advsimd && cpacr_asedis)
return disabledFault();
if ((cur_el == EL0 && cpacr_cp10 != 0x3) ||
(cur_el != EL0 && !(cpacr_cp10 & 0x1)))
return disabledFault();
}
if (fpexc_check && !fpexc.en)
return disabledFault();
// -- aarch32/exceptions/traps/AArch32.CheckFPAdvSIMDTrap --
if (have_virtualization && !is_secure && ELIs64(tc, EL2))
return checkFPAdvSIMDTrap64(tc, cpsr);
if (have_virtualization && !is_secure) {
HCPTR hcptr = tc->readMiscReg(MISCREG_HCPTR);
bool hcptr_cp10 = hcptr.tcp10;
bool hcptr_tase = hcptr.tase;
if (have_security && !ELIs64(tc, EL3) && !is_secure) {
if (nsacr.nsasedis)
hcptr_tase = true;
if (nsacr.cp10)
hcptr_cp10 = true;
}
if ((advsimd && hcptr_tase) || hcptr_cp10) {
const uint32_t iss = advsimd ? (1 << 5) : 0xA;
if (cur_el == EL2) {
return std::make_shared<UndefinedInstruction>(
machInst, iss,
EC_TRAPPED_HCPTR, mnemonic);
} else {
return std::make_shared<HypervisorTrap>(
machInst, iss,
EC_TRAPPED_HCPTR);
}
}
}
if (have_security && ELIs64(tc, EL3)) {
HCPTR cptrEnCheck = tc->readMiscReg(MISCREG_CPTR_EL3);
if (cptrEnCheck.tfp)
return advSIMDFPAccessTrap64(EL3);
}
return NoFault;
}
inline bool
ArmStaticInst::isWFxTrapping(ThreadContext *tc,
ExceptionLevel tgtEl,
bool isWfe) const
{
bool trap = false;
SCTLR sctlr = ((SCTLR)tc->readMiscReg(MISCREG_SCTLR_EL1));
HCR hcr = ((HCR)tc->readMiscReg(MISCREG_HCR_EL2));
SCR scr = ((SCR)tc->readMiscReg(MISCREG_SCR_EL3));
switch (tgtEl) {
case EL1:
trap = isWfe? !sctlr.ntwe : !sctlr.ntwi;
break;
case EL2:
trap = isWfe? hcr.twe : hcr.twi;
break;
case EL3:
trap = isWfe? scr.twe : scr.twi;
break;
default:
break;
}
return trap;
}
Fault
ArmStaticInst::checkForWFxTrap32(ThreadContext *tc,
ExceptionLevel targetEL,
bool isWfe) const
{
// Check if target exception level is implemented.
assert(ArmSystem::haveEL(tc, targetEL));
// Check for routing to AArch64: this happens if the
// target exception level (where the trap will be handled)
// is using aarch64
if (ELIs64(tc, targetEL)) {
return checkForWFxTrap64(tc, targetEL, isWfe);
}
// Check if processor needs to trap at selected exception level
bool trap = isWFxTrapping(tc, targetEL, isWfe);
if (trap) {
uint32_t iss = isWfe? 0x1E00001 : /* WFE Instruction syndrome */
0x1E00000; /* WFI Instruction syndrome */
switch (targetEL) {
case EL1:
return std::make_shared<UndefinedInstruction>(
machInst, iss,
EC_TRAPPED_WFI_WFE, mnemonic);
case EL2:
return std::make_shared<HypervisorTrap>(machInst, iss,
EC_TRAPPED_WFI_WFE);
case EL3:
return std::make_shared<SecureMonitorTrap>(machInst, iss,
EC_TRAPPED_WFI_WFE);
default:
panic("Unrecognized Exception Level: %d\n", targetEL);
}
}
return NoFault;
}
Fault
ArmStaticInst::checkForWFxTrap64(ThreadContext *tc,
ExceptionLevel targetEL,
bool isWfe) const
{
// Check if target exception level is implemented.
assert(ArmSystem::haveEL(tc, targetEL));
// Check if processor needs to trap at selected exception level
bool trap = isWFxTrapping(tc, targetEL, isWfe);
if (trap) {
uint32_t iss = isWfe? 0x1E00001 : /* WFE Instruction syndrome */
0x1E00000; /* WFI Instruction syndrome */
switch (targetEL) {
case EL1:
return std::make_shared<SupervisorTrap>(machInst, iss,
EC_TRAPPED_WFI_WFE);
case EL2:
return std::make_shared<HypervisorTrap>(machInst, iss,
EC_TRAPPED_WFI_WFE);
case EL3:
return std::make_shared<SecureMonitorTrap>(machInst, iss,
EC_TRAPPED_WFI_WFE);
default:
panic("Unrecognized Exception Level: %d\n", targetEL);
}
}
return NoFault;
}
Fault
ArmStaticInst::trapWFx(ThreadContext *tc,
CPSR cpsr, SCR scr,
bool isWfe) const
{
Fault fault = NoFault;
if (cpsr.el == EL0) {
fault = checkForWFxTrap32(tc, EL1, isWfe);
}
if ((fault == NoFault) &&
ArmSystem::haveEL(tc, EL2) && !inSecureState(scr, cpsr) &&
((cpsr.el == EL0) || (cpsr.el == EL1))) {
fault = checkForWFxTrap32(tc, EL2, isWfe);
}
if ((fault == NoFault) &&
ArmSystem::haveEL(tc, EL3) && cpsr.el != EL3) {
fault = checkForWFxTrap32(tc, EL3, isWfe);
}
return fault;
}
Fault
ArmStaticInst::checkSETENDEnabled(ThreadContext *tc, CPSR cpsr) const
{
bool setend_disabled(false);
ExceptionLevel pstateEL = (ExceptionLevel)(uint8_t)(cpsr.el);
if (pstateEL == EL2) {
setend_disabled = ((SCTLR)tc->readMiscRegNoEffect(MISCREG_HSCTLR)).sed;
} else {
// Please note: in the armarm pseudocode there is a distinction
// whether EL1 is aarch32 or aarch64:
// if ELUsingAArch32(EL1) then SCTLR.SED else SCTLR[].SED;
// Considering that SETEND is aarch32 only, ELUsingAArch32(EL1)
// will always be true (hence using SCTLR.SED) except for
// instruction executed at EL0, and with an AArch64 EL1.
// In this case SCTLR_EL1 will be used. In gem5 the register is
// mapped to SCTLR_ns. We can safely use SCTLR and choose the
// appropriate bank version.
// Get the index of the banked version of SCTLR:
// SCTLR_s or SCTLR_ns.
auto banked_sctlr = snsBankedIndex(
MISCREG_SCTLR, tc, !inSecureState(tc));
// SCTLR.SED bit is enabling/disabling the ue of SETEND instruction.
setend_disabled = ((SCTLR)tc->readMiscRegNoEffect(banked_sctlr)).sed;
}
return setend_disabled ? undefinedFault32(tc, pstateEL) :
NoFault;
}
Fault
ArmStaticInst::undefinedFault32(ThreadContext *tc,
ExceptionLevel pstateEL) const
{
// Even if we are running in aarch32, the fault might be dealt with in
// aarch64 ISA.
if (generalExceptionsToAArch64(tc, pstateEL)) {
return undefinedFault64(tc, pstateEL);
} else {
// Please note: according to the ARM ARM pseudocode we should handle
// the case when EL2 is aarch64 and HCR.TGE is 1 as well.
// However this case is already handled by the routeToHyp method in
// ArmFault class.
return std::make_shared<UndefinedInstruction>(
machInst, 0,
EC_UNKNOWN, mnemonic);
}
}
Fault
ArmStaticInst::undefinedFault64(ThreadContext *tc,
ExceptionLevel pstateEL) const
{
switch (pstateEL) {
case EL0:
case EL1:
return std::make_shared<SupervisorTrap>(machInst, 0, EC_UNKNOWN);
case EL2:
return std::make_shared<HypervisorTrap>(machInst, 0, EC_UNKNOWN);
case EL3:
return std::make_shared<SecureMonitorTrap>(machInst, 0, EC_UNKNOWN);
default:
panic("Unrecognized Exception Level: %d\n", pstateEL);
break;
}
return NoFault;
}
Fault
ArmStaticInst::sveAccessTrap(ExceptionLevel el) const
{
switch (el) {
case EL1:
return std::make_shared<SupervisorTrap>(machInst, 0, EC_TRAPPED_SVE);
case EL2:
return std::make_shared<HypervisorTrap>(machInst, 0, EC_TRAPPED_SVE);
case EL3:
return std::make_shared<SecureMonitorTrap>(machInst, 0,
EC_TRAPPED_SVE);
default:
panic("Illegal EL in sveAccessTrap\n");
}
}
Fault
ArmStaticInst::checkSveTrap(ThreadContext *tc, CPSR cpsr) const
{
const ExceptionLevel el = (ExceptionLevel) (uint8_t) cpsr.el;
if (ArmSystem::haveVirtualization(tc) && el <= EL2) {
CPTR cptrEnCheck = tc->readMiscReg(MISCREG_CPTR_EL2);
if (cptrEnCheck.tz)
return sveAccessTrap(EL2);
}
if (ArmSystem::haveSecurity(tc)) {
CPTR cptrEnCheck = tc->readMiscReg(MISCREG_CPTR_EL3);
if (!cptrEnCheck.ez)
return sveAccessTrap(EL3);
}
return NoFault;
}
Fault
ArmStaticInst::checkSveEnabled(ThreadContext *tc, CPSR cpsr, CPACR cpacr) const
{
const ExceptionLevel el = (ExceptionLevel) (uint8_t) cpsr.el;
if ((el == EL0 && cpacr.zen != 0x3) ||
(el == EL1 && !(cpacr.zen & 0x1)))
return sveAccessTrap(EL1);
return checkSveTrap(tc, cpsr);
}
static uint8_t
getRestoredITBits(ThreadContext *tc, CPSR spsr)
{
// See: shared/functions/system/RestoredITBits in the ARM ARM
const ExceptionLevel el = opModeToEL((OperatingMode) (uint8_t)spsr.mode);
const uint8_t it = itState(spsr);
if (!spsr.t || spsr.il)
return 0;
// The IT bits are forced to zero when they are set to a reserved
// value.
if (bits(it, 7, 4) != 0 && bits(it, 3, 0) == 0)
return 0;
const bool itd = el == EL2 ?
((SCTLR)tc->readMiscReg(MISCREG_HSCTLR)).itd :
((SCTLR)tc->readMiscReg(MISCREG_SCTLR)).itd;
// The IT bits are forced to zero when returning to A32 state, or
// when returning to an EL with the ITD bit set to 1, and the IT
// bits are describing a multi-instruction block.
if (itd && bits(it, 2, 0) != 0)
return 0;
return it;
}
static bool
illegalExceptionReturn(ThreadContext *tc, CPSR cpsr, CPSR spsr)
{
const OperatingMode mode = (OperatingMode) (uint8_t)spsr.mode;
if (unknownMode(mode))
return true;
const OperatingMode cur_mode = (OperatingMode) (uint8_t)cpsr.mode;
const ExceptionLevel target_el = opModeToEL(mode);
HCR hcr = ((HCR)tc->readMiscReg(MISCREG_HCR_EL2));
SCR scr = ((SCR)tc->readMiscReg(MISCREG_SCR_EL3));
if (target_el > opModeToEL(cur_mode))
return true;
if (!ArmSystem::haveEL(tc, target_el))
return true;
if (target_el == EL1 && ArmSystem::haveEL(tc, EL2) && scr.ns && hcr.tge)
return true;
if (target_el == EL2 && ArmSystem::haveEL(tc, EL3) && !scr.ns)
return true;
bool spsr_mode_is_aarch32 = (spsr.width == 1);
bool known, target_el_is_aarch32;
std::tie(known, target_el_is_aarch32) = ELUsingAArch32K(tc, target_el);
assert(known || (target_el == EL0 && ELIs64(tc, EL1)));
if (known && (spsr_mode_is_aarch32 != target_el_is_aarch32))
return true;
if (!spsr.width) {
// aarch64
if (!ArmSystem::highestELIs64(tc))
return true;
if (spsr & 0x2)
return true;
if (target_el == EL0 && spsr.sp)
return true;
} else {
// aarch32
return unknownMode32(mode);
}
return false;
}
CPSR
ArmStaticInst::getPSTATEFromPSR(ThreadContext *tc, CPSR cpsr, CPSR spsr) const
{
CPSR new_cpsr = 0;
// gem5 doesn't implement single-stepping, so force the SS bit to
// 0.
new_cpsr.ss = 0;
if (illegalExceptionReturn(tc, cpsr, spsr)) {
// If the SPSR specifies an illegal exception return,
// then PSTATE.{M, nRW, EL, SP} are unchanged and PSTATE.IL
// is set to 1.
new_cpsr.il = 1;
if (cpsr.width) {
new_cpsr.mode = cpsr.mode;
} else {
new_cpsr.width = cpsr.width;
new_cpsr.el = cpsr.el;
new_cpsr.sp = cpsr.sp;
}
} else {
new_cpsr.il = spsr.il;
if (spsr.width && unknownMode32((OperatingMode)(uint8_t)spsr.mode)) {
new_cpsr.il = 1;
} else if (spsr.width) {
new_cpsr.mode = spsr.mode;
} else {
new_cpsr.el = spsr.el;
new_cpsr.sp = spsr.sp;
}
}
new_cpsr.nz = spsr.nz;
new_cpsr.c = spsr.c;
new_cpsr.v = spsr.v;
if (new_cpsr.width) {
// aarch32
const ITSTATE it = getRestoredITBits(tc, spsr);
new_cpsr.q = spsr.q;
new_cpsr.ge = spsr.ge;
new_cpsr.e = spsr.e;
new_cpsr.aif = spsr.aif;
new_cpsr.t = spsr.t;
new_cpsr.it2 = it.top6;
new_cpsr.it1 = it.bottom2;
} else {
// aarch64
new_cpsr.daif = spsr.daif;
}
return new_cpsr;
}
bool
ArmStaticInst::generalExceptionsToAArch64(ThreadContext *tc,
ExceptionLevel pstateEL) const
{
// Returns TRUE if exceptions normally routed to EL1 are being handled
// at an Exception level using AArch64, because either EL1 is using
// AArch64 or TGE is in force and EL2 is using AArch64.
HCR hcr = ((HCR)tc->readMiscReg(MISCREG_HCR_EL2));
return (pstateEL == EL0 && !ELIs32(tc, EL1)) ||
(ArmSystem::haveEL(tc, EL2) && !inSecureState(tc) &&
!ELIs32(tc, EL2) && hcr.tge);
}
unsigned
ArmStaticInst::getCurSveVecLenInBits(ThreadContext *tc)
{
return tc->getIsaPtr()->getCurSveVecLenInBits(tc);
}
}